In an electronic power device, heating is an important consideration in relation to high power application over an extended period of time. The greater the operative energy, i.e., the product between the voltage across the device and the current flowing in it, the more energy the device will use for dissipating the generated heat to prevent its failure due to overheating.
The temperature reached by the device is usually identified as the junction temperature. For example, the critical junction temperature typically be greater than Timax.
Control circuits are used for thermal protection. They enable automatically turning off the power device when a critical temperature is reached, or when a current limit is detected.
Some control circuits require the integration of temperature sensors “on-chip” in the silicon die. If a threshold temperature is detected by one of the sensors, the device shuts down automatically, and the power device is kept turned off until the temperature detected by the sensors returns below a safety level by dissipating the heat through the package.
As schematically shown in FIG. 1, the sensors are arranged at the hottest points of the semiconductor die 50, and the detected temperatures are processed in parallel by the control circuits by an OR operator. The positions of the hottest points may already be known, for example, by a thermal map 51, as schematically shown in FIG. 2. FIG. 2 shows a test which the present Applicant executed on a power device by applying a load profile in a defined time range of 0.5 seconds repeated for 40 times.
FIG. 3 schematically shows a further possible arrangement of a power device 52 provided with sixteen channels and a corresponding number of sensors positioned at the single channels.
Prior approaches, even though they are suitable in some circumstances, may present some drawbacks. Indeed, during the power device operation, due to the high operative current density, reciprocal thermal impedances may be generated between adjacent channels. The temperatures detected by the sensors may be greater than the real temperatures, so that the power device may be turned off before reaching the temperature critical limits. This drawback may be substantially more acute in power devices with vertical dissipation.
Different causes inducing a turning off of the power device may include: the high dissipation of the electric power due to an overload caused by a short circuit, a wrong load control path, or the presence of inefficient solder connections; ambient temperature of the electric board that is too high due to the operation of other adjacent electronic devices; and the occurrence of an unexpected event.
An approach for avoiding such drawbacks is described in U.S. Pat. No. 5,008,736 to Davies et al., granted on 16, Apr. 1991, in which two overlapped chips, a first chip with the power transistor and second chip with the protection circuitry, are connected to obtain thermal protection. Such an approach, even though it may be suitable under certain circumstances, may not necessarily be generally extendable to other implementations.
Moreover, it may be desirable to avoid a complete turnoff of the power device if the critical events affect only a portion of the device.
It may also be desirable to detect currents just under the threshold which are damaging if they are endured for too much time.
It may further be desirable to define a thermal control process which may be effective also for devices operating with a very high current density, and with a number of junctions or channels in a range from 7 to 16, for example.
Further, it may be desirable to have a process which may be used with power devices operating with a junction critical temperature greater than about 175° C.